Can You Code Just by Speaking? A Hands‑Free Voice Guide to AI Programming Assistants

AI‑assisted coding has become popular, and many developers now want to eliminate repetitive typing by using voice commands. The author introduces a "voice programming" workflow that combines GitHub Copilot, Copilot Chat, the VS Code Speech extension, and a Chinese language pack to achieve hands‑free code generation.

GitHub Copilot and VS Code Speech

Copilot was the first to experiment with voice interaction through the now‑canceled Copilot Voice, but the functionality resurfaced via the VS Code Speech plugin. VS Code Speech provides bidirectional voice‑text conversion and integrates tightly with Copilot Chat, allowing users to invoke voice input in any chat interface and even hear replies spoken aloud.

Activating voice input uses the same shortcut as opening inline chat ( ⌘I or Ctrl+I ), so pressing the shortcut twice launches the chat and switches it to voice mode.

In the editor, the shortcuts ⌥⌘V (macOS) or Ctrl+Alt+V (Windows) trigger voice input, making it convenient for writing documentation or comments.

Installation

Four extensions are required:

GitHub Copilot – core coding assistant.

GitHub Copilot Chat – provides chat capabilities.

VS Code Speech – Microsoft’s voice plugin.

Chinese (Simplified, China) language support for VS Code Speech – adds Chinese speech recognition and synthesis.

Install them from VS Code’s Extensions panel, then open Settings (VS Code → Preferences → Settings) and search for "voice" to locate the relevant options.

The most important setting is Accessibility › Voice: Speech Language , which should be set to "中文(简体，中国)" so the Chinese voice pack is active.

The voice pack runs entirely locally; no network connection is needed for recognition or synthesis, resulting in fast and accurate performance.

Additional Configuration

Accessibility › Voice: Speech Timeout – defines how long a pause ends the voice input, eliminating the need to press Esc manually.

Accessibility › Voice: Keyword Activation – enables a wake‑word such as "Hey Code" to start voice input and choose the interaction mode (inline chat, quick chat, or chat panel).

Accessibility › Voice: Auto Synthesize – when enabled, the assistant replies audibly whenever the user asks a question by voice.

Using Voice with Cursor

Cursor, a VS Code‑based IDE, does not ship native voice support. Because it is compatible with VS Code extensions, the same VS Code Speech plugin and Chinese voice pack can be installed in Cursor. After configuration, the ⌥⌘V / Ctrl+Alt+V shortcuts activate dictation in the editor, which works well for large text blocks or comment‑based prompts.

However, voice input cannot be activated in certain UI elements such as inline chat ( ⌘K), Composer ( ⌘I), Notepad ( ⇧⌘I), or the chat panel, indicating limited compatibility.

System‑Level Speech Input

Both macOS and Windows provide built‑in speech input. On macOS, the author enables "Dictation" in System Preferences → Keyboard → Dictation, selects the Chinese (Mandarin – China) language, and uses the default "Hold Fn" shortcut to start and stop dictation. This method works in any Cursor input field, offering a viable fallback when the extension cannot be used.

Input‑Method Speech

Chinese input methods such as Sogou or iFlytek also offer high‑accuracy speech input. By configuring a convenient shortcut, these IME‑based voice inputs can match or exceed the system dictation experience.

Conclusion

Voice interaction upgrades the AI coding experience, allowing developers to code by speaking with GitHub Copilot, Cursor, and related tools. The guide demonstrates installation, configuration, and practical tips, encouraging readers to try the hands‑free workflow for a smoother and more enjoyable development process.